Current-Fed Resonant Full-Bridge Boost DC/AC/DC Converter
INDIRECT dc/ac/dc converters utilizing an isolation transformer
are widely used in different applications, such as
battery chargers and dischargers, uninterruptible power systems,
alternative energy systems, hybrid electric vehicles, and
medical X-ray imaging.
In the case of the applications where low input voltages have
to be converted to high output voltages, current-fed converters
are used, whereas in the case of higher power applications, fullbridge
boost converters are usually a good choice. The latter
type of converters will be dealt with in this paper.
The main advantage of such systems (in addition to having
a voltage gain greater than 1) is that they include a capacitive
output filter, which is preferred in higher voltage applications
[1]. Moreover, their output rectifier diodes are operated in a
discontinuous conduction mode with zero-current switching
(ZCS), which is an additional benefit.
One of the disadvantages of all-switched mode converters
concerns transformer parasitic elements. First, the transformer
leakage inductance causes undesirable voltage spikes that may
damage the circuit components, and second, the winding capacitance
may result in current spikes.
A vital factor that determines the size and the cost of a
converter is its operation frequency. In order to minimize the
size and the cost, the frequency has to be maximized. However,
higher frequencies result in the increase of transistor switching
losses, and thus, the converter’s effectiveness is limited. For
Manuscript received December 11, 2006; revised September 14, 2007. This
work was supported by the Science and Education Ministry under Project
W/WE/12/06.
The authors are with the Faculty of Electrical Engineering, Bialystok Technical
University, 15-351 Bialystok, Poland (e-mail: jasta[at]pb.edu.pl; tcitko@
pb.edu.pl).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TIE.2007.909084
that reason, many solutions have been proposed to minimize
converter switching losses.
The most popular control method of bridge converters is
pulsewidth modulation (PWM). It utilizes a phase-shift control
technique with constant frequency operation. The fixedfrequency
phase-shift control enables the implementation of
ZCS for all converter switches [2]. However, the switches
must provide a reverse-voltage blocking capability. Thus, they
have to be constructed by means of an insulated-gate bipolar
transistor (IGBT) or MOSFET in series with reverse-voltage
blocking diodes. The use of the diodes increases the component
count and cost, although it also causes higher conduction losses.
Another solution proposed to minimize converter switching
losses is a PWM boost full-bridge converter [3], in which the
leading switches realize ZCS under wide load range, and the
lagging switches realize zero-voltage switching (ZVS) under
any load. Likewise in this solution, the leading switches have
to be connected in series with reverse-voltage blocking diodes.
In addition, a circulating current, which is a result of the
introduction of an additional auxiliary inductance (connected
parallel with the primary winding of the transformer), is the
source of extra conduction losses. Moreover, the switching
frequency is limited by the use of IGBTs as the leading switches
(10 kHz).
The next solution to reduce converter switching losses that
is presented in the literature [4]–[6] introduces an active-clamp
network. It utilizes the resonance between a clamp capacitor
and a transformer leakage inductance in order to limit bridge
switch turn-off voltage overshoot and to enable the use of
energy stored in the leakage inductance for ZVS. However, this
solution is also not devoid of several drawbacks, as discussed
in [4]. First, in order to avoid current flow between the transformer
parasitic capacitance and inductance during a boost
inductor charge cycle, series blocking diodes should be added.
As a result, extra conduction loss appears, and the possibility
of obtaining ZVS for the bridge transistors is lost. Additionally,
due to a parasitic capacitance, ZCS for the lagging transistors
may not be possible at high line currents.
Another interesting topology that is presented in the literature
is a current-fed zero-voltage transition PWM converter [7]–[9].
Due to the use of an auxiliary network, all the power switches
are zero-voltage switched. However, the auxiliary network requires
an additional transistor that needs to be controlled, and
this results in additional power losses.
A separate group of dc/ac/dc converters, which is different
from PWM solutions, is frequency-controlled resonant converters.
In these converters, the input current is not directly controlled
but depends on the resonant circuit impedance and the
0278-0046/$25.00 © 2008 IEEE
JALBRZYKOWSKI AND CITKO: CURRENT-FED RESONANT FULL-BRIDGE BOOST DC/AC/DC CONVERTER 1199
load value. Among the converters are series–parallel resonant
converters using LCC-type commutation [10]–[13], which are
preferred for higher voltage applications. These converters,
which are controlled above the resonant frequency, are characterized
by the voltage gain greater than 1. The range of the
voltage gain change grows with the increase of the quality
factor value.
There are several advantages to the operation above the resonant
frequency; among them are the elimination of transistor
turn-on losses, operation with slow recovery feedback diodes,
and the possibility for the introduction of lossless snubbers (i.e.,
capacitors connected across switches) to eliminate transistor
turn-off losses.
The characteristic feature of resonant converters is that the
transformer parasites do not disturb the circuit, because they
are used as resonant circuit elements. Nevertheless, at least
three disadvantages of such systems can be mentioned. First,
the voltage gain is highly sensitive to the quality factor (load resistance).
Second, the relative increase of the transistor current
caused by the increase of the quality factor (load resistance)
reduces the system’s efficiency. Finally, the system requires an
output inductive filter, which is not preferred in high voltage
dc/dc converters.
Another solution in the group of frequency-controlled resonant
converters is a current-fed parallel resonant push–pull
inverter [14], which is used as a power supply for ozone
generators. The resonant circuit comprises a transformer magnetizing
inductance and ozonizer capacitances. Although this
application uses a dc/ac converter topology with a relatively low
operation frequency of 5–7 kHz (thus, switching losses do not
present any problems), transistor switch-off voltage overshoot
still remains as its main disadvantage.
Finally, the last solution to be mentioned, and the most
similar to the solution that will be proposed in this paper, is a
current-fed full-bridge boost converter [1]. Because of its control
method, this converter may be classified as a quasi-resonant
converter. In this type of control method, the overlapping
conduction time of the four converter switches is kept constant
and the output voltage is regulated by varying the switching
frequency. The conduction time is particularly calculated to
ensure ZCS operation under a wide load range. MOSFETs and
body diodes are used as the converter switches without the need
for any additional diodes in series.
As far as the drawbacks of this solution are concerned, the
control of the output voltage is achieved through decreasing
frequency, which, with a large control range, results in a low
frequency.
Another common problem for both of these, as well as
all the aforementioned solutions, concerns MOSFET parasitic
capacitance, which, when operating with different inductances,
may introduce parasitic oscillations during the time when the
transistor is open.
The converter proposed in this paper has the advantages
of all the aforementioned solutions and, at the same time,
is devoid of their drawbacks. The converter transistor turnoff
time is constant and is equal to the time of the parallelconnected
capacitor overcharge. The output voltage is regulated
by varying the switching frequency. Due to this control tech-
Fig. 1. Proposed converter circuit. (a) Main circuit configuration. (b) Transistor
control pulse waveforms.
nique, the converter MOSFETs are switched in zero-voltage
conditions.
An additional characteristic feature of the proposed control
technique is that, for the nominal operation point, all the
transistor control pulses overlap during a well-defined part
of a period. Thus, the converter is boost-derived. Moreover,
the output voltage is regulated from the nominal operation
point by a frequency increase, which involves a simultaneous
decrease of the control pulse overlap. In spite of the fact that the
converter is boost-derived, its transistors do not require reversevoltage
blocking in series, which is its other advantage.
Taking into account the aforementioned characteristics of
the proposed converter, it can be classified as a multiresonant
converter with a series–parallel resonant circuit but with a
completely new feature, i.e., it is boost-derived.
For the analysis of the proposed converter as a nonlinear system,
the first harmonic method was implemented. The method,
which is widely used for the description of class E converters
[15], assumes that the bridge converter output current is sinusoidal
and its input current is constant, and thus the method
does not divide the whole control period in particular modes of
operation.
This paper first presents the analysis of the system operation
that is supported by a mathematical description that is useful for
the system design. Next, it provides PSPICE model simulations,
as well as experimental results, for a 1.4-kW prototype, which
is controlled with 250–370 kHz.
II. CONVERTER SYSTEM DESCRIPTION
The proposed converter scheme is shown in Fig. 1(a). The
inductance Ll represents the transformer leakage inductance,
the capacitance C includes the transformer parasitic capacitance,
and the capacitances CT include the transistor parasitic
capacitances.
The transistor control pulse waveforms are shown in
Fig. 1(b). The converter is controlled by varying the switching
frequency f = 1/T , while simultaneously keeping a constant
1200 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 55, NO. 3, MARCH 2008
break between the pulses. Thus, the transistor pulse overlap tov
gradually decreases while the switching frequency increases.
The maximal output voltage (power) is achieved with the minimal
switching frequency, which is marked as nominal, i.e., fn.
For this frequency, the converter operates in the optimal operation
point while its transistors are switched at zero-voltage
and zero-current conditions. This optimal operation point is
achieved by assuming the constant break between the control
pulses depending on the resonant circuit elements and the load
resistance nominal (minimal) value. In the optimal operation
point, converter currents alternatively flow through transistors
or its external capacitors CT . The transistor body diodes do not
participate in current conduction. During one control period,
the following subintervals can be determined: during the pulse
overlap time tov, all the transistors are on, and because of
the system symmetry, each of them conducts a half value of
the input current, i.e., Iin/2, and a half value of the output
current, i.e., iL/2. During the constant break between the
control pulses, two transistors and two capacitors alternatively
conduct T1T4CT3CT2 or T2T3CT1CT4. Because of the system
symmetry, each transistor and each capacitor conducts, as
in the previous subinterval, half values of the input and output
currents. Thus, for the system mathematical analysis, there is
no need to divide the entire switching cycle into particular
subintervals.
When the switching frequency increases above its nominal
value or the load resistance increases above its nominal value,
the converter goes to a suboptimal operation range. In the
suboptimal operation state, the constant break between the
control pulses is too long for the capacitors CT to overcharge.
Thus, after the capacitor overcharge, an adequate transistor
body diode takes over current conduction. After the body diode
current falls to zero, it starts to increase in the transistor at zerovoltage
conditions.
The main task of the system design is the calculation of the
transistor pulse overlap tov and the constant break between the
control pulses in accordance with the resonant circuit element
parameters, i.e., C, L, and CT , and the load resistance for the
assumed value of the nominal switching frequency, in order to
achieve the converter optimal operation point.